Dissertations / Theses on the topic 'Uncertain sequences'

Lo scopo di questa tesi è di elaborare e analizzare metodi rigorosi dal punto di vista matematico per l’analisi di due tipi di dati biologici: dati relativi a pan-genomi e filogenesi. Con il termine “pan-genoma” si indica, in generale, un insieme di sequenze genomiche strettamente correlate (tipicamente appartenenti a individui della stessa specie) che si vogliano utilizzare congiuntamente come sequenze di riferimento per un’intera popolazione. Una filogenesi, invece, rappresenta le relazioni evolutive in un gruppo di entità, che siano esseri viventi, geni, lingue naturali, manoscritti antichi o cellule tumorali. Con l’eccezione di uno dei risultati presentati in questa tesi, relativo all’analisi di filogenesi tumorali, il taglio della dissertazione è prevalentemente teorico: lo scopo è studiare gli aspetti combinatori dei problemi affrontati, più che fornire soluzioni efficaci in pratica. Una conoscenza approfondita degli aspetti teorici di un problema, del resto, permette un'analisi matematicamente rigorosa delle soluzioni già esistenti, individuandone i punti deboli e quelli di forza, fornendo preziosi dettagli sul loro funzionamento e aiutando a decidere quali problemi vadano ulteriormente investigati. Oltretutto, è spesso il caso che nuovi risultati teorici (algoritmi, strutture dati o riduzioni ad altri problemi più noti) si possano direttamente applicare o adattare come soluzione ad un problema pratico, o come minimo servano ad ispirare lo sviluppo di nuovi metodi efficaci in pratica. La prima parte della tesi è dedicata a nuovi metodi per eseguire delle operazioni fondamentali su un testo elastico-degenerato, un oggetto computazionale che codifica in maniera compatta un insieme di testi simili tra loro, come, ad esempio, un pan-genoma. Nello specifico, si affrontano il problema di cercare una sequenza di lettere in un testo elastico-degenerato, sia in maniera esatta che tollerando un numero prefissato di errori, e quello di confrontare due testi degenerati. Nella seconda parte si considerano sia filogenesi tumorali, che ricostruiscono per l'appunto l'evoluzione di un tumore, sia filogenesi "classiche", che rappresentano, ad esempio, la storia evolutiva delle specie viventi. In particolare, si presentano nuove tecniche per confrontare due o più filogenesi tumorali, necessarie per valutare i risultati di diversi metodi che ricostruiscono le filogenesi stesse, e una nuova e più efficiente soluzione a un problema di lunga data relativo a filogenesi "classiche", consistente nel determinare se sia possibile sistemare, in presenza di dati mancanti, un insieme di specie in un albero filogenetico che abbia determinate proprietà.
The main goal of this thesis is to develop new algorithmic frameworks to deal with (i) a convenient representation of a set of similar genomes and (ii) phylogenetic data, with particular attention to the increasingly accurate tumor phylogenies. A “pan-genome” is, in general, any collection of genomic sequences to be analyzed jointly or to be used as a reference for a population. A phylogeny, in turn, is meant to describe the evolutionary relationships among a group of items, be they species of living beings, genes, natural languages, ancient manuscripts or cancer cells. With the exception of one of the results included in this thesis, related to the analysis of tumor phylogenies, the focus of the whole work is mainly theoretical, the intent being to lay firm algorithmic foundations for the problems by investigating their combinatorial aspects, rather than to provide practical tools for attacking them. Deep theoretical insights on the problems allow a rigorous analysis of existing methods, identifying their strong and weak points, providing details on how they perform and helping to decide which problems need to be further addressed. In addition, it is often the case where new theoretical results (algorithms, data structures and reductions to other well-studied problems) can either be directly applied or adapted to fit the model of a practical problem, or at least they serve as inspiration for developing new practical tools. The first part of this thesis is devoted to methods for handling an elastic-degenerate text, a computational object that compactly encodes a collection of similar texts, like a pan-genome. Specifically, we attack the problem of matching a sequence in an elastic-degenerate text, both exactly and allowing a certain amount of errors, and the problem of comparing two degenerate texts. In the second part we consider both tumor phylogenies, describing the evolution of a tumor, and “classical” phylogenies, representing, for instance, the evolutionary history of the living beings. In particular, we present new techniques to compare two or more tumor phylogenies, needed to evaluate the results of different inference methods, and we give a new, efficient solution to a longstanding problem on “classical” phylogenies: to decide whether, in the presence of missing data, it is possible to arrange a set of species in a phylogenetic tree that enjoys specific properties.

Earthquake statistics is a growing field of research with direct application to probabilistic seismic hazard evaluation. The earthquake process is a complex spatio-temporal phenomenon, and has been thought to be an example of the self-organised criticality (SOC) paradigm, in which events occur as cascades on a wide range of sizes, each determined by fine details of the rupture process. As a consequence, deterministic prediction of specific event sizes, locations, and times may well continue to remain elusive. However, probabilistic forecasting, based on statistical patterns of occurrence, is a much more realistic goal at present, and is being actively explored and tested in global initiatives. This thesis focuses on the temporal statistics of earthquake populations, exploring the uncertainties in various commonly-used procedures for characterising seismicity and explaining the origins of these uncertainties. Unlike many other SOC systems, earthquakes cluster in time and space through aftershock triggering. A key point in the thesis is to show that the earthquake inter-event time distribution is fundamentally bimodal: it is a superposition of a gamma component from correlated (co-triggered) events and an exponential component from independent events. Volcano-tectonic earthquakes at Italian and Hawaiian volcanoes exhibit a similar bimodality, which in this case, may arise as the sum of contributions from accelerating and decelerating rates of events preceding and succeeding volcanic activity. Many authors, motivated by universality in the scaling laws of critical point systems, have sought to demonstrate a universal data collapse in the form of a gamma distribution, but I show how this gamma form is instead an emergent property of the crossover between the two components. The relative size of these two components depends on how the data is selected, so there is no universal form. The mean earthquake rate—or, equivalently, inter-event time—for a given region takes time to converge to an accurate value, and it is important to characterise this sampling uncertainty. As a result of temporal clustering and non-independence of events, the convergence is found to be much slower than the Gaussian rate of the central limit theorem. The rate of this convergence varies systematically with the spatial extent of the region under consideration: the larger the region, the closer to Gaussian convergence. This can be understood in terms of the increasing independence of the inter-event times with increasing region size as aftershock sequences overlap in time to a greater extent. On the other hand, within this high-overlap regime, a maximum likelihood inversion of parameters for an epidemic-type statistical model suffers from lower accuracy and a systematic bias; specifically, the background rate is overestimated. This is because the effect of temporal overlapping is to mask the correlations and make the time series look more like a Poisson process of independent events. This is an important result with practical relevance to studies using inversions, for example, to infer temporal variations in background rate for time-dependent hazard estimation.

Sequence alignment is one of the most intensely studied problems in bioinformatics, and is an important step in a wide range of analyses. An issue that has gained much attention in recent years is the fact that downstream analyses are often highly sensitive to the specific choice of alignment. One way to address this is to jointly sample alignments along with other parameters of interest. In order to extend the range of applicability of this approach, the first chapter of this thesis introduces a probabilistic evolutionary model for protein structures on a phylogenetic tree; since protein structures typically diverge much more slowly than sequences, this allows for more reliable detection of remote homologies, improving the accuracy of the resulting alignments and trees, and reducing sensitivity of the results to the choice of dataset. In order to carry out inference under such a model, a number of new Markov chain Monte Carlo approaches are developed, allowing for more efficient convergence and mixing on the high-dimensional parameter space. The second part of the thesis presents a directed acyclic graph (DAG)-based approach for representing a collection of sampled alignments. This DAG representation allows the initial collection of samples to be used to generate a larger set of alignments under the same approximate distribution, enabling posterior alignment probabilities to be estimated reliably from a reasonable number of samples. If desired, summary alignments can then be generated as maximum-weight paths through the DAG, under various types of loss or scoring functions. The acyclic nature of the graph also permits various other types of algorithms to be easily adapted to operate on the entire set of alignments in the DAG. In the final part of this work, methodology is introduced for alignment-DAG-based sequence annotation using hidden Markov models, and RNA secondary structure prediction using stochastic context-free grammars. Results on test datasets indicate that the additional information contained within the DAG allows for improved predictions, resulting in substantial gains over simply analysing a set of alignments one by one.

In single-cell analysis, the physiologic states of individual cells are studied. In some studies, the subject of interest is the development over time of some cell characteristic. To obtain time-resolved single-cell data, one possibility is to conduct an experiment on a cell population and make a sequence of images of the population over the course of the experiment. If a mapping is at hand, which determines which cell it is that is the cause of each measured cell in the image sequence, time resolved single-cell data can be extracted. Such a mapping is called a lineage tree, and the process of creating it is called tracking. One aim of this work is to develop a tracking algorithm that incorporates organism specific knowledge, such as average division time, in the tracking process. With respect to this aim, a Bayesian model that incorporates biological knowledge is derived, with which every hypothetical lineage tree can be assigned a probability. Additionally, two Monte Carlo algorithms are developed, that approximate the probability distribution of lineage trees given by the Bayesian model. When an approximate distribution is known, for example the most likely lineage tree can be extracted and used. In many cases, the information provided to an automatic tracking algorithm is insufficient for the algorithm to find the gold standard lineage tree. In these cases, a possibility is to construct the gold standard lineage tree by manual correction of the lineage tree that has been provided by the tracking algorithm. A second aim of this work is to provide a confidence to every assignment in a lineage tree, in order to give the person doing manual corrections useful information about what assignments to change. Such a confidence is provided by the Monte Carlo tracking methods developed in this work.
I enskild-cell analys studeras de fysiologiska tillståndet hos enskilda celler. I vissa studier är man intresserad av hur någon cellegenskap utvecklas över tid. Ett sätt att generera tidsupplöst data på enskild-cellnivå är att utföra ett experiment med en cellpopulation och avbilda den med mikroskop med jämna mellanrum. Med hjälp av en avbildning som beskriver vilken cell i experiment det är som ger upphov till vilken uppmätt cell i bildsekvensen, kan sedan enskild-cell data tillgås. En sådan avbildning kallas ett stamträd (lineage tree), och processen att bestämma stamträdet kallas tracking. En målsättning med detta arbete är att utveckla en trackingalgoritm som använder organismspecifik kunskap, såsom organismens genomsnittliga delningstid, i trackingprocessen. Med denna målsättning i hänseende härleds en bayesiansk modell med vilken varje stamträd kan tillskrivas en sannolikhet, och som kan ta hänsyn till biologisk fakta när detta sker. Därtill utvecklas två Monte Carlo algoritmer som approximerar sannolikhetsfördelningen av stamträd som härrör ur den bayesianska modellen. När en uppskattad fördelning är känd kan t ex det mest sannolika stamträdet i fördelningen användas för enskild-cell analys. I många fall är informationen som en automatisk trackingalgoritm har till hands inte tillräcklig för att algoritmen ska kunna producera gold standard stamträdet. I dessa fall kan det vara befogat att konstruera gold standard stamträdet genom att göra manuella korrektioner på ett stamträd som tagits fram automatiskt med en algoritm. En andra målsättning med detta arbete är att införa ett förtroendemått för enskilda länkar i ett stamträd. Detta förtroendemått ska göra det enklare för personen som gör manuella korrektioner att avgöra ifall en länk i ett stamträd behöver korrigeras eller ej. Ett sådant förtroendemått införs, och de två Monte Carlo algoritmerna som utvecklas i detta arbete tillskriver ett förtroende för varje länk i de stamträd som de levererar.

Abstract The purpose of this research has been to show with applications that polyhedral scenes can be modeled in real time with a single video camera. Sometimes this can be done very efficiently without any special image processing hardware. The developed vision sensor estimates its three-dimensional position with respect to the environment and models it simultaneously. Estimates become recursively more accurate when objects are approached and observed from different viewpoints. The modeling process starts by extracting interesting tokens, like lines and corners, from the first image. Those features are then tracked in subsequent image frames. Also some previously taught patterns can be used in tracking. A few features in the same image are extracted. By this way the processing can be done at a video frame rate. New features appearing can also be added to the environment structure. Kalman filtering is used in estimation. The parameters in motion estimation are location and orientation and their first derivates. The environment is considered a rigid object in respect to the camera. The environment structure consists of 3-D coordinates of the tracked features. The initial model lacks depth information. The relational depth is obtained by utilizing facts such as closer points move faster on the image plane than more distant ones during translational motion. Additional information is needed to obtain absolute coordinates. Special attention has been paid to modeling uncertainties. Measurements with high uncertainty get less weight when updating the motion and environment model. The rigidity assumption is utilized by using shapes of a thin pencil for initial model structure uncertainties. By observing continuously motion uncertainties, the performance of the modeler can be monitored. In contrast to the usual solution, the estimations are done in separate state vectors, which allows motion and 3-D structure to be estimated asynchronously. In addition to having a more distributed solution, this technique provides an efficient failure detection mechanism. Several trackers can estimate motion simultaneously, and only those with the most confident estimates are allowed to update the common environment model. Tests showed that motion with six degrees of freedom can be estimated in an unknown environment. The 3-D structure of the environment is estimated simultaneously. The achieved accuracies were millimeters at a distance of 1-2 meters, when simple toy-scenes and more demanding industrial pallet scenes were used in tests. This is enough to manipulate objects when the modeler is used to offer visual feedback.

xi, 119 p. : ill. (some col.)
The evolutionary history of protein families can be difficult to study because necessary ancestral molecules are often unavailable for direct observation. As an alternative, the field of computational phylogenetics has developed statistical methods to infer the evolutionary relationships among extant molecular sequences and their ancestral sequences. Typically, the methods of computational phylogenetic inference and ancestral sequence reconstruction are combined with other non-computational techniques in a larger analysis pipeline to study the inferred forms and functions of ancient molecules. Two big problems surrounding this analysis pipeline are computational error and statistical uncertainty. In this dissertation, I use simulations and analysis of empirical systems to show that phylogenetic error can be reduced by using an alternative search heuristic. I then use similar methods to reveal the relationship between phylogenetic uncertainty and the accuracy of ancestral sequence reconstruction. Finally, I provide a case-study of a molecular machine in yeast, to demonstrate all stages of the analysis pipeline. This dissertation includes previously published co-authored material.
Committee in charge: John Conery， Chair； Daniel Lowd， Member； Sara Douglas， Member； Joseph W. Thornton， Outside Member

Natural disasters can create significant uncertainty for individuals and entire cities. This thesis examines the role of government decision-making and uncertainty in disaster recovery, focusing on a case study of post-earthquake Christchurch, New Zealand. Beginning in September 2010, Christchurch has been shaken by a devastating sequence of earthquakes, stretching over 18 months. The most severe event took place on February 22, 2011, taking the lives of 185 people and causing significant damage throughout the city. Building damage has forced the closure of portions of the Central Business District (CBD) for over 2 years as of July 2013, and over 7,000 residential properties have been purchased by the government due to land damage. The duration of the earthquake sequence, combined with the scale of damage, has created significant uncertainty for the city, specifically for the future of the CBD and the local property market. This thesis seeks to examine how government decision-making can incentivize a community of self-interested actors facing uncertainty to pull together, and create an outcome that benefits all of them. A conceptual framework is developed through which three key government decisions in the Christchurch case are analyzed in terms of how uncertainty has been managed. The three decisions are: 1) maintaining a Cordon around the CBD, 2) Establishing the Christchurch Central Development Unit to plan the rebuild of the CBD, and 3) Establishing a system of zoning to classify land damage for residential properties. A detailed description of the earthquake sequence and context is also provided. The primary research for this thesis was collected during 23 semi-structured key informant interviews conducted in New Zealand in May of 2012. Interviewees were selected with expertise in a range of different recovery issues, as well as different roles in the recovery, from decision-makers to those implementing the decisions, and those impacted. In conclusion, this thesis argues that uncertainty has been a major driver in government decision-making, and that those decisions have had a significant impact in terms of reducing uncertainty. In particular, decisions have addressed uncertainty in terms of the residential property market, and the future of the CBD.

A multiple sequence alignment (MSA) provides a description of the relationship between biological sequences where columns represent a shared ancestry through an implied set of evolutionary events. The majority of research in the field has focused on improving the accuracy of alignments within the progressive alignment framework and has allowed for powerful inferences including phylogenetic reconstruction, homology modelling and disease prediction. Notwithstanding this, when applied to modern genomics datasets - often comprising tens of thousands of sequences - new challenges arise in the construction of accurate MSA. These issues can be generalised to form three basic problems. Foremost, as the number of sequences increases, progressive alignment methodologies exhibit a dramatic decrease in alignment accuracy. Additionally, for any given dataset many possible MSA solutions exist, a problem which is exacerbated with an increasing number of sequences due to alignment uncertainty. Finally, technical difficulties hamper the deployment of such genomic analysis workflows - especially in a reproducible manner - often presenting a high barrier for even skilled practitioners. This work aims to address this trifecta of problems through a web server for fast homology extension based MSA, two new methods for improved phylogenetic bootstrap supports incorporating alignment uncertainty, a novel alignment procedure that improves large scale alignments termed regressive MSA and finally a workflow framework that enables the deployment of large scale reproducible analyses across clusters and clouds titled Nextflow. Together, this work can be seen to provide both conceptual and technical advances which deliver substantial improvements to existing MSA methods and the resulting inferences.
Un alineament de seqüència múltiple (MSA) proporciona una descripció de la relació entre seqüències biològiques on les columnes representen una ascendència compartida a través d'un conjunt implicat d'esdeveniments evolutius. La majoria de la investigació en el camp s'ha centrat a millorar la precisió dels alineaments dins del marc d'alineació progressiva i ha permès inferències poderoses, incloent-hi la reconstrucció filogenètica, el modelatge d'homologia i la predicció de malalties. Malgrat això, quan s'aplica als conjunts de dades de genòmica moderns, que sovint comprenen desenes de milers de seqüències, sorgeixen nous reptes en la construcció d'un MSA precís. Aquests problemes es poden generalitzar per formar tres problemes bàsics. En primer lloc, a mesura que augmenta el nombre de seqüències, les metodologies d'alineació progressiva presenten una disminució espectacular de la precisió de l'alineació. A més, per a un conjunt de dades, existeixen molts MSA com a possibles solucions un problema que s'agreuja amb un nombre creixent de seqüències a causa de la incertesa d'alineació. Finalment, les dificultats tècniques obstaculitzen el desplegament d'aquests fluxos de treball d'anàlisi genòmica, especialment de manera reproduïble, sovint presenten una gran barrera per als professionals fins i tot qualificats. Aquest treball té com a objectiu abordar aquesta trifecta de problemes a través d'un servidor web per a l'extensió ràpida d'homologia basada en MSA, dos nous mètodes per a la millora de l'arrencada filogenètica permeten incorporar incertesa d'alineació, un nou procediment d'alineació que millora els alineaments a gran escala anomenat MSA regressivu i, finalment, un marc de flux de treball permet el desplegament d'anàlisis reproduïbles a gran escala a través de clústers i computació al núvol anomenat Nextflow. En conjunt, es pot veure que aquest treball proporciona tant avanços conceptuals com tècniques que proporcionen millores substancials als mètodes MSA existents i les conseqüències resultants.

Most evolutionary analyses are based upon pre-estimated multiple sequence alignment models. From a computational point of view, it is too complex to estimate a correct alignment, as it is to derive a correct tree from that alignment. Several works have recently reported on the influence of alignment on downstream analysis, and on the uncertainty inherent to their estimation. Chapter 1 develops the notion of alignment uncertainty as either inherent to the data (internal) or resulting from methodological biases (external). Chapter 2 presents two contributions of mine for the improvement of MSA methods through the use of homology extension (TM-Coffee) and thanks to an improved word-matching algorithm (SymAlign). In Chapter 3, I show how alignment uncertainty can be used to improve the trustworthiness of phylogenetic analysis. Chapter 4 shows how a similar improvement can be obtained through a simple adaptation of the T-Coffee transitive score, thus allowing downstream analysis to take into account internal alignment uncertainty. The final chapter contained a discussion of our current results and possible future work.
La mayoría de los análisis evolutivos están basados en modelos establecidos de alineamiento de secuencia múltiple. Desde un punto de vista computacional, es igual de complejo la estimación de un alineamiento correcto, como la obtención de un árbol correcto a partir del alineamiento. Recientemente varios trabajos han informado sobre la influencia del alineamiento en los análisis posteriores, y en la incertidumbre inherente a su estimación. El Capítulo 1 desarrolla el concepto de incertidumbre de alineación, tanto inherente a los datos (internos), como resultante de los sesgos metodológicos (externo). El Capítulo 2 presenta dos contribuciones mías para la mejora de los métodos de MSA a través del uso de la extensión de homología (TM‐Coffee) y gracias a un algoritmo de coincidencia de palabra mejorado (SymAlign). En el capítulo 3, se muestra cómo la incertidumbre de alineación puede ser utilizada para mejorar la confiabilidad del análisis filogenético. El capítulo 4 nos muestra como se puede obtener una mejora similar por medio de una simple adaptación de la puntuación transitiva del T-- Coffee, lo cual permite un análisis posterior para tener en cuenta la incertidumbre de alineación interna. El último capítulo contiene un análisis de los resultados actuales y los posibles futuros trabajos.

This thesis examines the response of a representative agent investor to sequences (streaks) of quarterly earnings surprises over a period of twelve quarters using the United States S&P500 constituent companies sample frame in the years 1991 to 2006. This examination follows the predictive performance of the representative agent model of Rabin (2002b) [Inference by believers in the law of small numbers. The Quarterly Journal of Economics. 117(3).p.775 816] and Barberis, Shleifer, and Vishny (1998) [A model of investor sentiment. Journal of Financial Economics. 49. p.307 343] for an investor who might be under the influence of the law of small numbers, or another closely related cognitive bias known as the gambler s fallacy. Chapters 4 and 5 present two related empirical studies on this broad theme. In chapter 4, for successive sequences of annualised quarterly earnings changes over a twelve-quarter horizon of quarterly earnings increases or falls, I ask whether the models can capture the likelihood of reversion. Secondly, I ask, what is the representative investor s response to observed sequences of quarterly earnings changes for my S&P500 constituent sample companies? I find a far greater frequency of extreme persistent quarterly earnings rises (of nine quarters and more) than falls and hence a more muted reaction to their occurrence from the market. Extreme cases of persistent quarterly earnings falls are far less common than extreme rises and are more salient in their impact on stock prices. I find evidence suggesting that information discreteness; that is the frequency with which small information about stock value filters into the market is one of the factors that foment earnings momentum in stocks. However, information discreteness does not subsume the impact of sequences of annualised quarterly earnings changes, or earnings streakiness as a strong candidate that drives earnings momentum in stock returns in my S&P500 constituent stock sample. Therefore, earnings streakiness and informational discreteness appear to have separate and additive effects in driving momentum in stock price. In chapter 5, the case for the informativeness of the streaks of earnings surprises is further strengthened. This is done by examining the explanatory power of streaks of earnings surprises in a shorter horizon of three days around the period when the effect of the nature of earnings news is most intense in the stock market. Even in shorter windows, investors in S&P500 companies seem to be influenced by the lengthening of negative and positive streaks of earnings surprises over the twelve quarters of quarterly earnings announcement I study here. This further supports my thesis that investors underreact to sequences of changes in their expectations about stock returns. This impact is further strengthened by high information uncertainties in streaks of positive earnings surprise. However, earnings streakiness is one discrete and separable element in the resolution of uncertainty around equity value for S&P 500 constituent companies. Most of the proxies for earnings surprise show this behaviour especially when market capitalisation, age and cash flow act as proxies of information uncertainty. The influence of the gambler s fallacy on the representative investor in the presence of information uncertainty becomes more pronounced when I examine increasing lengths of streaks of earnings surprises. The presence of post earnings announcement drift in my large capitalised S&P500 constituents sample firms confirms earnings momentum to be a pervasive phenomenon which cuts across different tiers of the stock markets including highly liquid stocks, followed by many analysts, which most large funds would hold.

Le travail produit s'inscrit dans un cadre industriel piloté par la société 2MoRO Solutions. La réalisation présentée dans cette thèse doit servir à l'élaboration d'un service à haute valeur, permettant aux exploitants aéronautiques d'optimiser leurs actions de maintenance. Les résultats obtenus permettent d'intégrer et de regrouper les tâches de maintenance en vue de minimiser la durée d'immobilisation des aéronefs et d'en réduire les risques de panne.La méthode que nous proposons comporte trois étapes : (i) une étape de rationalisation des séquences afin de pouvoir les combiner [...]
The present work is part of an industrial project driven by 2MoRO Solutions company.It aims to develop a high value service enabling aircraft operators to optimize their maintenance actions.Given the large amount of data available around aircraft exploitation, we aim to analyse the historical events recorded with each aircraft in order to extract maintenance forecasting. Theresults are used to integrate and consolidate maintenance tasks in order to minimize aircraft downtime and risk of failure. The proposed method involves three steps : (i) streamlining information in order to combinethem, (ii) organizing this data for easy analysis and (iii) an extraction step of useful knowledgein the form of interesting sequences. [...]

Today, the main research field for the automotive industry is to find solutions for active safety. In order to perceive the surrounding environment, tracking nearby traffic objects plays an important role. Validation of the tracking performance is often done in staged traffic scenarios, where additional sensors, mounted on the vehicles, are used to obtain their true positions and velocities. The difficulty of evaluating the tracking performance complicates its development. An alternative approach studied in this thesis, is to record sequences and use non-causal algorithms, such as smoothing, instead of filtering to estimate the true target states. With this method, validation data for online, causal, target tracking algorithms can be obtained for all traffic scenarios without the need of extra sensors. We investigate how non-causal algorithms affects the target tracking performance using multiple sensors and dynamic models of different complexity. This is done to evaluate real-time methods against estimates obtained from non-causal filtering. Two different measurement units, a monocular camera and a LIDAR sensor, and two dynamic models are evaluated and compared using both causal and non-causal methods. The system is tested in two single object scenarios where ground truth is available and in three multi object scenarios without ground truth. Results from the two single object scenarios shows that tracking using only a monocular camera performs poorly since it is unable to measure the distance to objects. Here, a complementary LIDAR sensor improves the tracking performance significantly. The dynamic models are shown to have a small impact on the tracking performance, while the non-causal application gives a distinct improvement when tracking objects at large distances. Since the sequence can be reversed, the non-causal estimates are propagated from more certain states when the target is closer to the ego vehicle. For multiple object tracking, we find that correct associations between measurements and tracks are crucial for improving the tracking performance with non-causal algorithms.

碩士
大葉大學
電機工程學系碩士班
93
Floorplanning is an important issue in the physical design phase of the VLSI design. As VLSI design hierarchy becomes more complex, it is difficult to immediately send obtained floorplanning information back to frontend system level designer for modifying modules if floorplanning is considered until the backend physical design phase. Hence, we consider solving floorplanning of uncertain modules which are designed in frontend system level and have not been designed completely yet. In this paper, to evaluate chip area effectively and efficiently, a non-slicing uncertain floorplan algorithm based on clustering strategy is proposed. In the method, clustering strategy is first applied to group modules to form fewer supermodules of larger size, then the Sequence-Pair representation for non-slicing floorplan is used to record the relative positions among supermodules, and the simulated annealing procedure is finally executed according to Sequence-Pair for obtaining an area distributed graph. Experimental results on uncertain floorplanning demonstrate the efficiency and effectiveness of proposed algorithm.

碩士
國立中興大學
電機工程學系所
101
Sequence data are ubiquitous in our daily life, such as animals’ seasonal migration, DNA/protein sequences, Web browsing sequences. Sequence pattern mining is to discover special, important, and representative features hidden in sequence data. It attracts a lot of attention especially in the domains of bioinformatics and spatio-temporal trajectory data mining. Sequence data are inherently of some uncertainty, and the uncertainty may be caused by many reasons, such as limitations of the measuring technology, sampling error, privacy preserving. In this thesis, we focus on the mining of uncertain sequences to discover hidden patterns by using Probabilistic Suffix Tree (PST). PST is an implementation of Variable-length Markov Model (VMM) that is wildly used in sequence pattern mining in many domains. However, traditional PST building algorithm is designed to mine certain data and inapplicable of mining uncertain sequences. In addition, traditional PST building algorithm is a centralized algorithm such that it is incapable of handling huge amounts of accumulated uncertain data. Therefore, we propose a distributed and parallel algorithm on Hadoop platform to fully utilize the computing power of cloud computing in the uncertain sequence mining. In the thesis, we propose two distributed and parallel PST building algorithms, named CloudPST+ and CloudPST+_OneScan respectively on the Hadoop platform to speed up the learning process. Specifically, CloudPST+ is of Map/Reduce framework that builds a PST in an iterative and levels by levels manner to avoid learning excessive patterns and trade off the overhead of distributed computing. CloudPST+_OneScan extends CloudPST+ and involves a new data structure to store the intermediate statistics so that the One-Scan algorithm only scans the entire sequence data once in each iteration. To evaluate the performance of CloudPST+ and CloudPST+_OneScan, we implement a naïve approach derived from the well-known Wordcount example of Hadoop/MapReduce and conduct several experiments with a synthetic dataset that is re-generated from a real trajectory dataset. According to our experimental results, CloudPST+ and CloudPST+_OneScan significantly outperform the naïve approach. In addition, sacrificing an extra memory cost, CloudPST+_OneScan shows better efficiency, scalability, and stability than CloudPST+.